一作ICML 2026Poster· General Machine Learning / Evaluation
Beyond the Final Answer: Evaluating the Reasoning Trajectories of Tool-Augmented Agents
Wonjoong Kim · Sangwu Park · Yeonjun In · Sein Kim · Dongha Lee · Chanyoung Park
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Wonjoong Kim · Sangwu Park · Yeonjun In · Sein Kim · Dongha Lee · Chanyoung Park
Yeonjun In · Wonjoong Kim · Sangwu Park · Chanyoung Park
Large reasoning models (LRMs) achieve strong performance on complex reasoning tasks but often generate harmful responses to malicious user queries. This paper investigates the underlying cause of these safety risks and shows that the issue lies in the reasoning structure itself. Based on this ins...